The most important challenge in knock detection is to detect its intensity. Depending on the phenomenon characteristic the spark ignition calibration can be optimized. For this reason, the scope of this paper is the use the Discrete Wavelet Transform (DWT) as a tool to analyze knock signals characteristics in the time-scale decomposition. A brief description of the Short-Time Fourier Transform (STFT) analysis and comparisons between Fourier analysis and DWT are also shown.
Time-frequency analysis methods have become more usual in recent years and can be applied in different areas of the automotive field, such as noise, vibration and powertrain calibration. Due to the demand vehicles with better performance, fuel economy and emissions; the signal analysis tools have been important to optimize the system functionality, such as driveability and knocking. The knock is an undesired phenomenon and it is generated by the shock of flame fronts in the combustion chamber. The excessive cylinder peak pressure and fuel low octane number are the possible root causes that lead to knock occurrence. The phenomenon is detected by the presence of instability in the cylinder pressure curve of the spark-ignition engine, which can be measured through the vibration on the engine block.
The knock signal is considered as a non-stationary event (time-varying) and its main characteristics are the short period of time and the contained of high frequencies. However, these characteristics are often the most important part of the signal and Fourier analysis is not suited to detect the signal intensity. The Wavelet analysis allows the use of both long time intervals where we want more precise low-frequency information, and shorter regions where we want to emphasize high-frequency information and seems to be a good candidate to characterize the knocking phenomena.